Explaining Radical Group Behaviour:
Developing Emotion and Efficacy Routes to Normative and Non-normative Collective Action
Nicole Tausch1, Julia C. Becker2, Russell Spears3, Oliver Christ2, Rim Saab3, Purnima Singh4 and Roomana N Siddiqui5
1
University of St Andrews, Scotland 2
University of Marburg, Germany 3
Cardiff University, Wales 4
Indian Institute of Technology Delhi, India 5
Aligarh Muslim University, India
Abstract
A recent model of collective action distinguishes two distinct pathways; an emotional pathway whereby anger in response to injustice motivates action, and an efficacy pathway where the belief that issues can be solved collectively increases the likelihood that group members take action (van Zomeren, Spears, Fischer, & Leach, 2004). Research supporting this model has, however, focused entirely on relatively normative actions such as participating in
demonstrations. The authors argue that the relations between emotions, efficacy and action differ for more extreme, non-normative actions and propose (1) that non-normative actions are often driven by a sense of low efficacy and (2) that contempt, which, unlike anger, entails
psychological distancing and a lack of reconciliatory intentions, predicts non-normative action. These ideas are tested in three survey studies examining student protests against tuition fees in Germany (N = 332), Indian Muslims’ action support in relation to ingroup disadvantage (N = 156), and British Muslims’ responses to British foreign policy (N = 466). Results were generally supportive of predictions and indicated that (a) anger was strongly related to normative action but overall unrelated or less strongly related to non-normative action; (b) contempt was either unrelated or negatively related to normative action but significantly positively predicted non-normative action; and (c) that efficacy was positively related to non-normative action and negatively to non-normative action. The implications of these findings for understanding and dealing with extreme intergroup phenomena such as terrorism are discussed.
What is common to East Germans taking to the streets in 1989 to demand democratic reforms, British factory workers staging sit-in protests against planned redundancies, environmental activists ‘spiking’ trees to sabotage attempts to cut them down, and Palestinians bombing Israeli nightclubs in their struggle for national liberation is that people are acting on behalf of a group in order to achieve a group goal. According to Wright, Taylor and Moghaddam (1990a), “a group member engages in collective action any time that she or he is acting as a representative of the group and the action is directed at improving the conditions of the entire group” (p. 995). Such action is often aimed at challenging group-based discrimination or group disadvantage, or at ending or preventing an injustice. As the examples above illustrate, collective action can take on many forms, ranging from relatively moderate and non-violent actions like taking part in
peaceful demonstrations, signing petitions, or participating in acts of civil disobedience, to more radical forms such as sabotage, violence and terrorism.
What mobilizes people to engage in collective action has been a key question in the social sciences (e.g., Blumer, 1939; Gurr, 1970, 1993; Klandermans, 1997) and the central focus of prominent psychological theories of group behaviour such as Relative Deprivation Theory (RDT; e.g., Runciman, 1966; Walker & Smith, 2002) and Social Identity Theory (SIT; Tajfel & Turner, 1979). A vast amount of empirical research has examined the structural and
different action strategies is, however, essential for further theoretical development in the field (see Wright, 2009). Such work would also speak to recent efforts in psychological science to understand extreme intergroup phenomena such as terrorism (e.g., Reich, 1990; Victoroff & Kruglanski, 2009) and could afford vital information for practitioners and policy makers concerned with steering political action away from violent confrontation towards non-violent forms of engagement (Schwarzmantel, 2010).
To address this gap in the literature, the present research systematically examines predictors of different forms of collective action. Specifically, we distinguish normative (i.e., action that conforms to the norms of the wider social system) from non-normative action (i.e., action that violates these rules1; Wright et al., 1990a) and utilize a recent integrative theoretical model that proposes two distinct pathways to collective action; an emotional pathway whereby anger in response to injustice motivates action, and an efficacy pathway where the belief that issues can be solved collectively increases the likelihood that group members take action (van Zomeren, Spears, Fischer, & Leach, 2004). We argue that the relations between emotion, efficacy and action differ for more extreme, normative actions. First, we propose that non-normative actions are often driven by a sense of low (rather than high) efficacy. Second, rather than being driven primarily by anger, we propose that the experience of contempt, which entails psychological distance from its object and a lack of reconciliatory intentions (see Fischer & Roseman, 2007), predicts non-normative action. Before outlining our hypotheses in more detail, we summarize the work on the roles of emotion and efficacy in collective action.
A common starting point of psychological approaches to collective action is that people respond to a sense of disadvantage, unjust treatment, or threat. RDT for example posits that people engage in collective action as a result of viewing their group as relatively deprived or disadvantaged in comparison with a reference group (e.g., Guimond & Dube-Simard, 1983; Runciman, 1966). Whether collective action occurs further depends on interpretations of the social structure (Tajfel & Turner, 1979). Both RDT and SIT view issues of legitimacy and justice as central in this process and stress that ingroup disadvantage must be seen as illegitimate, unfair, or unjust (e.g., Ellemers, Wilke, & van Knippenberg, 1993; Mummendey, Kessler, Klink, & Mielke, 1999). Work on RDT further stresses that feelings of deprivation, such as anger, resentment, and outrage are important in driving action (see Walker & Smith, 2002).
This focus on emotional reactions is in line with recent work on Intergroup Emotion Theory (IET; Mackie, Devos & Smith, 2000; E.R. Smith, 1993). This approach is based on appraisal theories of emotion (e.g., Frijda, Kuipers, & te Schure, 1989), which view emotion as a complex ‘syndrome’ that involves cognitions, subjective feelings, and behavioural tendencies. Using insights from Self-Categorization Theory (SCT, Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), IET posits that, in situations where individuals categorize as members of a social group, group-related events become self-relevant and arouse emotions together with their associated action tendencies (E.R. Smith, 1993). Thus, the appraisal that the ingroup has been treated unfairly or suffered an unjust disadvantage arouses feelings of (group-based) anger and evokes action tendencies to move against the offender. Consistent with this view, there is considerable empirical support for the link between injustice appraisals and anger (see Miller, 2000; Weiss, Suckow, & Cropananzo, 1999), for the role of anger-related emotions in
Mackie et al., 2000; Pennekamp, Doosje, Zebel, & Fischer, 2007), as well as for the mediating role of anger in the relation between group-based appraisals and confrontational action
tendencies (e.g., Mackie et al., 2000; van Zomeren et al., 2004). The Role of Group Efficacy
Another line of research has put greater emphasis on pragmatic considerations and has highlighted the need for members of disadvantaged groups to believe that their group’s position is changeable (Gamson, 1992; Klandermans, 1997; Martin et al., 1984). This idea is related to the notion of stability in SIT (Tajfel & Turner, 1979), which posits that the perceived instability of the group hierarchy is necessary for collective action to occur. While stability refers to the situational constraints for change, concepts such as agency (Gamson, 1992) or collective efficacy (Bandura, 2000) correspond more closely to the extent to which the ingroup is viewed as being capable of solving their problems (see Mummendey et al., 1999). Wright (2001) combined these factors in the concept of collective control, which results from both the belief that the intergroup context is responsive to action and from the perception that the ingroup has the abilities to effect change. The importance of instrumental factors has also been the focus of resource mobilization theories (e.g., Klandermans, 1997; McCarthy & Zald, 1977), which emphasize the presence of necessary resources as a key determinant of collective action. Indeed, in its most radical form, this approach argues that, if people have the resources for effective mobilization, they will engage in collective action irrespective of whether they feel that they have been unjustly treated (e.g., McCarthy & Zald, 1977; see also Martin et al., 1984). Consistent with this general
Emotion and Efficacy as Dual Pathways to Collective Action
Rather than viewing these different accounts as competing explanations, a recent dual-pathway model conceptualizes emotion and efficacy perceptions as two distinct but
complementary pathways to collective action. Van Zomeren et al. (2004) likened collective action through these pathways to emotion- and problem-focused coping with collective disadvantage, respectively (cf. Lazarus, 1991). They provided empirical evidence across three experimental studies showing that anger, which resulted from experiences of procedural injustice and opinion support from other group members, and perceived group efficacy, which was
predicted by instrumental social support, independently predict collective action tendencies and thus constitute two separate explanatory pathways. Based on this integrative model, the present work examines emotion and efficacy perceptions as distinct predictors of group-based action.
The Present Research
The present research extends this previous work by considering a wider array of collective actions. How to classify different forms of action has been the subject of much
controversy in the literature on political engagement (Sabucedo & Arce, 1991). For the purposes of the present study, we apply Wright et al.’s (1990a) well-known distinction between normative (i.e., action that conforms to the norms of the existing social system, such as political
Historical analyses of political campaigns suggest that groups sometimes use both types of action over the course of their existence (Stephan & Chenoweth, 2008), or in parallel (e.g., the ‘ArmaLite and ballot box’ strategies of the Irish Republican Army, see Hayes and McAllister, 2005). Radical subgroups have also at times developed out of wider social movements, such as the Red Army Faction which emerged from the West German student protest movement (see Aust, 2008). Moskalenko and McCauley (2009) have suggested that, at the individual level, normative vs. non-normative action constitute two independent dimensions of political action that follow from different sets of appraisals of the political situation. Supporting this view, they demonstrated that normative and non-normative action form two correlated but clearly separate factors (see also Corning & Myers, 2002). In this paper, we aim to expand our understanding of the appraisals and emotions underlying these different forms of collective action. Extending van Zomeren et al.’s (2004) dual pathway model, we examine two key ideas relating to these paths:
Anger and Contempt as Predictors of Normative and Non-normative Collective Action First, we propose that anger should be related to normative but not non-normative action and that contempt would predict non-normative action. This argument follows from work on the functional differences between anger and contempt. Although anger and contempt are strongly related, co-occurring emotions that both imply negative appraisals of others’ intentions of (Frijda et al., 1989) and are ‘other-hostile’ (Izard, 1971), there are important differences between these emotions in terms of their development and implications for social relations (Fischer &
are aimed at changing the other person’s behaviour and will ultimately result in an improvement of the relationship (see also Averill, 1983). This is consistent with an approach to emotion which views anger as a constructive emotion that functions to correct wrongdoing and uphold accepted standards of conduct (see Averill, 1983; Weber, 2004).
Fischer and Roseman (2007) demonstrated that contempt, on the other hand, occurs in less intimate relationships, where there is a perceived lack of control over the other person, and where reconciliation is no longer sought. They showed that contempt can occur in response to the same instances of behavior as anger does, but often develops on top of anger; that is, it results from prior incidents of anger with the same person that went unresolved. Contempt is associated with permanent changes in the beliefs about another person and negative dispositional
attributions of the offending behavior (Fischer & Roseman, 2007). In the context of intergroup relations, contempt was also shown to be related to outgroup dehumanization (Esses, Veenvliet, Hodson, & Mihic, 2008), which can legitimate extreme actions (see Staub, 1990). Contempt often results in derogation of the object of contempt, the deterioration of social relationships, and social exclusion (Fischer & Roseman, 2007). Thus, contempt leads to a (physical and
psychological) distancing from the object of contempt.
extreme, non-normative forms of action (Hypothesis 2). While contempt can result in actions that serve to remove the object of contempt from one’s environment and is often associated with withdrawal (Fisher & Roseman, 2007; Mackie et al., 2000), it has also been suggested that contempt can be associated with ‘move against’ action tendencies (see E.R. Smith, 1993). Given that feelings of contempt are associated with a lack of reconciliation intention, dehumanization, and moral exclusion, we propose that, in the presence of an injustice or a threat, contempt can result in more extreme actions against an offender. This possibility was also acknowledged by Fischer and Roseman (2007) who suggested that the presence of contempt may lead to
particularly hostile reactions because attack tendencies are accompanied by extreme derogation and are not held in check by a desire to preserve social relationships. A psychological distancing from, and moral exclusion of, the object of contempt (which, in the context of political action could be the government, police, or an offending outgroup more generally) further undermines the need to adhere to social norms and moral standards when dealing with the offender. Given that non-normative action challenges the legitimacy of the current social system (see Wright, 2010) and seeks radical social change and reorganization, contempt should play a key role in predicting such action. Thus, we predict contempt to be uniquely and positively related to non-normative action tendencies (Hypothesis 3). Moreover, based on appraisal theories of emotion (Frijda et al., 1989; E.R. Smith, 1993), we expect both anger and contempt to be predicted by injustice appraisals, and injustice appraisals to be indirectly related to normative and non-normative action via anger and contempt, respectively (Hypotheses 4a and 4b).
Group Efficacy as a Predictor of Normative and Non-normative Action
collective action (see van Zomeren et al., 2008). As mentioned above, this evidence is, however, primarily based on studies examining relatively moderate, normative forms of action. We predict that while efficacy might be positively related to normative action (Hypothesis 5), it is likely to be negatively related to non-normative action (Hypothesis 6). That is, non-normative action should occur when individuals feel that their group is powerless to address an injustice or influence relevant political decisions. This might be because individuals feel that their group does not have access to the conventional channels of political influence (e.g., Wright, 2009; Wright et al., 1990a), is marginalized by the procedures of the existing political system (Gurr, 1993; Schwarzmantel, 2010), or is too disorganized or unsupportive of the cause (see van Zomeren et al., 2004). Low efficacy can also ensue because the government (or other relevant powerful group) is unresponsive to (Bandura, 2000; Wright, 2001), or even oppressive of (Drury & Reicher, 2005; Reicher, 2004), attempts to change the situation.
A number of findings are consistent with the idea that low efficacy drives more extreme forms of action. For example, Ransford (1968) demonstrated in the context of the Watts Riots in the United States that feelings of powerlessness and lack of control over events were positively correlated with willingness to engage in violence. Furthermore, Wright and colleagues showed in an experimental study that non-normative action was chosen when movement from a
disadvantaged group to an advantaged group was completely closed (Wright et al., 1990a). These authors also demonstrated that lack of hope for an improvement of their position best
Spears, Scheepers, & van Zomeren, 2010). Scheepers et al. referred to this as a ‘nothing to lose’ strategy, reasoning that the low status group had little to lose by reacting in a more provocative way, given that their situation was unlikely to change by doing nothing.
It should be noted that we do not propose that non-normative action represents an irrational strategy. Rather, we suggest that non-normative action can be highly strategic and serve a number of functions, such as influencing wider public opinion, building a movement, and winning third parties for the cause (e.g., Hornsey et al., 2006); for example by provoking the opponent into extreme counter-action (see Sedgwick, 2004). Thus, although the group as a whole might currently be seen as powerless and ineffective, non-normative action might unsettle the current political situation and thereby facilitate the conditions that could lead to social change in the long run (see also Louis & Taylor, 2002; Spears et al., 2010).
Overview of Studies
Because it is difficult to assess actual collective action, in particular if it is
non-normative, like most past research (see van Zomeren et al., 2008) we used either own willingness to engage in action or support for different forms of action as dependent measures in the present research. Previous research has shown that behavioural intentions can be a useful proxy for actual behaviour (Webb & Sheeran, 2006) and that collective action intentions are good predictors of actual participation (e.g., De Weerd & Klandermans, 1999; Moskalenko & McCauley, 2009). Furthermore, attitudes towards different forms of action are important
variables in their own right with considerable substantive importance. This has been illustrated in the terrorism literature, which suggests that public opinion plays an important role in actual terrorist activity (e.g., Krueger & Malečková, 2009; Mascini, 2006).
We test our hypotheses in three diverse contexts and in relation to a range of different criterion variables. Study 1 was conducted in the context of student protests against tuition fees in Germany and examined injustice appraisals, anger, contempt and efficacy perceptions as predictors of students’ willingness to engage in normative and non-normative collective action. Study 2 examined our predictions in a different cultural context, namely Indian Muslims’ support for actions in relation to ingroup disadvantage. Study 3 was conducted in the context of British Muslims’ responses to British foreign policy towards Muslim countries. This study included a wider range of criteria (voting intentions, normative collective action intentions, and support for violence against military and civilian targets) and used an index of political efficacy rather than general group efficacy as a predictor.
The right to a cost-free university education has long been a given in German society until the German high court decided to overturn a ban on tuition fees in 2005. Each state can now decide whether its universities will charge for their services, and how much. The court ruling and plans to introduce fees were met with outrage and nationwide demonstrations by students. The present research was conducted in the context of student protests in the federal state of Hessen, where tuition fees were introduced in October 2006. Students in Hessen took a number of actions to oppose tuition fees, ranging from relatively normative actions such as participating in
demonstrations and signing petitions, to more radical, non-normative, and illegal actions. These included non-violent actions such as blocking university buildings and highways and disturbing events and lectures, but also violent actions by a minority of students who set fires, destroyed property and attacked police (see Der Spiegel, 2007). The present study was conducted in January 2008, when a law suit against the constitutionality of tuition fees was underway and the future of tuition fees in Hessen was uncertain.
Method
Participants and Procedure
Data were collected as part of an online survey posted to various email lists at several universities in Hessen, Germany. The survey was completed by 332 students (162 female, 146 male, 24 unknown; mean age = 22.79, SD = 3.37) from a range of subjects. Items were presented in German. Upon completion, respondents were able to enter into a prize draw.
Measures Injustice appraisals
legitimate“; “The introduction of tuition fees is justified” (reverse-coded). Respondents indicated their agreement with these items using 7-point scales (1 = strongly disagree, 7 = strongly agree). Efficacy
Group efficacy was measured by four items (α = .84): “I think that students can stop the introduction of tuition fees”; “I think that students can successfully defend their rights”,
“Students are strong as a group and can move a lot”; and “I think students have already lost the fight against tuition fees” (reverse coded). Respondents indicated their agreement with these items using 7-point scales (1 = strongly disagree, 7 = strongly agree).
Anger
To assess anger, respondents indicated their agreement with the items “I’m furious about the planned introduction of tuition fees” and “The introduction of tuition fees angers me” (1 = strongly disagree, 7 = strongly agree; r = .88, p <.001).
Contempt
Contempt was assessed using two items: “I disdain people who advocate tuition fees” and “I detest people who advocate tuition fees” (1 = strongly disagree, 7 = strongly agree). The items were averaged to form an index of contempt (r = .87, p <.001).
Action tendencies
Respondents were asked to indicate how likely it is that they would participate in 16 different actions against tuition fees in the future, all of which had occurred as part of the student protest in the past (1 = very unlikely, 7 = very likely). Principal components analysis yielded three components with Eigenvalues greater than 1 which accounted for 77.44 % of the variance. Loadings, after oblique rotation, showed that relatively normative actions (participate in
unconstitutionality of tuition fees, street theatre, demonstrations) loaded primarily on the first component (>.55), clearly non-normative, violent actions (throw stones or bottles, arson attacks on university buildings, arson attacks on private property of responsible persons, attacks on police, attacks on responsible persons) loaded on the second component (>.89), and more moderate, non-violent non-normative actions (disturb events where advocates of tuition fees appear, block university buildings, block the highway) loaded on the third component (>.74). These three components were also identified in a previous study by Sabucedo and Arce (1991), who distinguished political participation that operates within the political system from violent and non-violent action that operates outside of the system. Two items (boycott tuition fees, go on strike) that had cross-loadings on both the normative and the non-violent, non-normative
component were excluded from the analyses. The remaining items were averaged to yield
composites of own likelihood to engage in normative action (α = .90), non-violent non-normative actions (α = .89), and violent non-normative action (α = .96).
Results and Discussion
Means, standard deviations, and zero-order correlations are presented in Table 1. Preliminary Analyses and General Analytic Strategy
may not be so in the population. One approach to handling nonnormal data is the bootstrap method (West, Finch, & Curran, 1995), a resampling technique whereby multiple samples are drawn randomly, with replacement, from the original sample resulting in a bootstrap sampling distribution from which standard errors and confidence intervals are calculated (see Efron & Tibshirani, 1993). Although the bootstrap sampling distribution operates in the same way as the sampling distribution in parametric inferential statistics, it is free from assumptions of normality (Byrne, 2009). Moreover, bootstrap standard errors and confidence intervals for indirect effects are superior compared to standard ways of estimating standard errors of indirect effects like the Sobel Test (Preacher & Hayes, 2008; Shrout & Bolger, 2002). Given the advantages of this technique, we decided to use bootstrapping for our main regression analyses in all studies. In all studies, our regression analyses treated action intentions (or support) as the criterion variables, injustice and efficacy perceptions as predictors and emotions (anger and contempt) as mediators. We also controlled for age and gender (all studies) and SES (Studies 2 and 3)2. We used Mplus 5.2 (Muthén & Muthén, 1998-2007) to be able to estimate all relevant model parameters (total, direct, and indirect effects; cf. Preacher & Hayes, 2008) in one step. Complete results of all analyses are presented in Tables S1-S3 in the supplementary materials.
bootstrapping improves results of FIML considerably, even under strong deviations from normality.
Main Analysis: Injustice Perceptions, Efficacy, and Emotions as Predictors of Action Tendencies Consistent with previous research, appraisals of injustice of tuition fees significantly predicted feelings of anger (B = .82, SE = .04, p < .001; CI: .748/.902). Injustice perceptions were also positively related to feelings of contempt (B = .37, SE = .05, p < .001; CI: .277/.469). Anger predicted willingness to engage in normative (B = .38, SE = .08, p < .001; CI: .220/.531) and non-violent non-normative (B = .23, SE = .08, p = .006; CI: .060/.379) action, but was unrelated to violent non-normative action (B = .00, SE = .01, p = .742; CI: -.020/.012).
Contempt, on the other hand, was unrelated to willingness to engage in normative (B = -.04, SE = .05, p = .452; CI: .136/.057) and nonviolent nonnormative (B = .11, SE = .07, p = .127; CI: -.029/.248) action, but significantly predicted willingness to engage in violent non-normative action (B = .02, SE = .01, p = .036; CI: .002/.034). The indirect effects of injustice appraisals on willingness to engage in normative (B = .32, SE = .06, p < .001; CI: .186/.436) and non-violent non-normative (B = .29, SE = .07, p = .005; CI: .050/.314) action via anger, and the indirect effect of injustice appraisals on violent non-normative action via contempt (B = .01, SE = .003, p = .058; CI: .001/.0143), were significant. There was also a significant direct relation between injustice appraisals and normative action (B = .20, SE = .08, p = .011; CI: .051/.360).
To provide a stronger test of the idea that anger and contempt differentially predict normative and non-normative action, we also tested whether the differences in predictive power of anger and contempt for the different forms of action were significant. We did this by
comparing a model where the paths in question were constrained to be equal with an
whether differences were significant. Because we distinguished 3 different types of action, we first compared a model that constrained the two non-normative actions to be equal with the unconstrained model. For the relation between contempt and action tendencies, there was no difference between this constrained model and the unconstrained model (∆χ2 = 2.32, df = 1, p = .13), indicating that the relation between contempt and these two forms of non-normative action were comparable. We then tested whether this model differed from a more restrictive model that constrained all three paths from contempt to action to be equal. This more restrictive model differed significantly from the less restrictive model (∆χ2 = 3.93, df = 1, p < .05), indicating that contempt more strongly predicts non-normative action than normative action.
We repeated these tests for the paths from anger to action tendencies. The model that constrained the two non-normative actions to be equal differed significantly from the
unconstrained model (∆χ2 = 7.88, df = 1, p < .01), indicating that anger had significantly different relations to the two forms of non-normative action. We therefore conducted pair wise
comparisons between paths, which indicated that the relations between anger and action differed significantly between all types of action (all ps <.05), such that anger was significantly less predictive of action the more extreme the action was.
Perceived group efficacy was, as expected, positively related to willingness to engage in normative action (B = .36, SE = .06, p < .001; CI: .249/.460). The relation between efficacy and non-violent, non-normative action was positive and approached significance (B = .12, SE = .07, p = .079; CI: -.018/.242). Consistent with our hypothesis, group efficacy was significantly
To summarize, these findings present first evidence that different appraisals and emotions underlie different forms of collective action. This was primarily apparent when comparing normative and violent non-normative actions. As expected, anger was related to normative but not to violent non-normative action. Additional tests comparing the relative strength of paths further indicated that anger was significantly less predictive the more extreme the action. Thus, it seems that for anger there was a continuous diminution of predictive power as a function of extremity of criterion action. Overall, these findings are in line with current thinking that anger is a constructive emotion which is likely to result in actions that are bound to conventional norms and allow for reconciliation (Averill, 1983; Fischer & Roseman, 2007). The present study further provided evidence for our hypothesis that contempt would predict non-normative forms of action. This link between contempt and non-normative action was only significant for violent non-normative action. Thus, contempt, which is often associated with dehumanization and moral exclusion of the object of contempt and a lack of reconciliatory intentions (see Fischer &
Roseman, 2007), may contribute particularly toward extreme collective actions in the context of a group-based injustice.
that, in the present context, actions such as blocking streets and buildings were seen as legitimate and fairly normative and acceptable strategies as many students engaged in these activities.
Our next study examines our hypotheses further, this time in a different cultural context and an environment of enduring inequality and violent intergroup conflict, and in relation to a different set of criterion variables. This study was conducted among Muslims in India.
Study 2
Muslims constitute India’s largest religious minority of about 13.4 percent of the total population (80.5 percent of which is Hindu; Census of India, 2001). They have suffered from economic and social disadvantage since India’s partition in 1947 and are now among the most disadvantaged communities in the country in terms of education, income, employment and political representation (Sachar Committee Report, 2006). Unlike for other disadvantaged groups, such as ‘scheduled’ caste Hindus, there are no targeted efforts by the government to improve conditions for Muslims. Furthermore, violent riots between Hindu and Muslim communities have plagued India with regularity since partition, costing tens of thousands of lives. While violence is often incited by Hindu nationalist groups, local inequalities and economic competition between communities have undoubtedly played a role (see Singh, 1988, for a review). The present study was conducted at Aligarh Muslim University (AMU) in Aligarh, Uttar Pradesh, where the majority of students and staff are Muslim. Muslims are, however, a minority in the town of Aligarh and the university-town relationship has been conflictual. In fact, Aligarh is one of the most riot-prone cities in India (see Varshney, 2002).
government policies that would address Muslim disadvantage and their attitudes towards the use of violence by Muslim groups. It should be noted that while support for political violence clearly represents non-normative action as examined in the previous study, support for government policies differs conceptually from the measures of normative action used in Study 1. The measures of normative action in the previous study represented confrontational actions (actions aimed at forcing the government to change policies) using means within the existing political system. Support for government policies is, however, not confrontational and does not represent collective action as such. It therefore is beyond the scope of the dual pathway model and we would not expect anger and ingroup efficacy to predict this attitude. Nonetheless, we decided to include this measure as a criterion variable in order to explore whether the same variables that foster non-normative action might also predict a psychological distancing from the dominant system (i.e., less endorsement of actions proposed or taken by the dominant group).
Method
Participants and Procedure
The survey was administered as a paper-and-pencil questionnaire. Participants were recruited during classes at several departments at AMU. A total of 169 students who
self-identified as Muslims participated. Twelve participants were excluded from our analyses because they had missing values for several complete scales. Our final sample thus comprised 157
participants (82 female, 74 male, 1 unknown) with a mean age of 21.66 (SD = 2.06). The questionnaire was administered in English which is widely understood and spoken at AMU. Among a number of additional measures pertaining to Hindu-Muslim relations in India, the questionnaire included items assessing our key variables.
Perceived disadvantage
To assess perceived disadvantage of the ingroup, participants indicated the extent to which they agreed with the item “I often think that Hindus are favoured and Muslims disadvantaged in India” (1 = strongly disagree, 5 = strongly agree).
Efficacy
Group efficacy was measured by two items (r = .39, p <.001). On scales ranging from 1 (strongly disagree) to 5 (strongly agree), respondents indicated their agreement with the items “I think that Muslims as a group are able to improve their situation” and “Muslims can together overcome their difficulties”.
Anger
On scales ranging from 1 (not at all) to 5 (extremely) respondents indicated the extent to which they felt angry, resentful, furious, and displeased when thinking about the disadvantaged status of Muslims in India (α = .86).
Contempt
Respondents indicated the extent to which they felt contemptuous when thinking about the disadvantaged status of Muslims in India (1 = not at all; 5 = extremely).
Policy support
Support for violence
Our measure of support for ingroup violence was adopted from Hayes and McAllister (2005). Respondents indicated their agreement with three items (α = .83): “In general, I
understand some Muslim groups’ reasons for the use of violence, even though I do not condone the violence itself”; “In general, I have sympathy for some Muslim groups’ reasons to resort to violent means in general, even though I do not condone the violence itself”; “In general, I support some Muslim groups’ decisions to use violence, even though I do not condone the violence itself” (1 = strongly disagree, 5 = strongly agree).
Results and Discussion
Means, standard deviations, and zero-order correlations between key variables are presented in Table 2. Because there was a small amount of scattered missing data (≤ 5.7%), we again used FIML estimation. Perceived disadvantage of the ingroup significantly predicted both feelings of anger (B = .26, SE = .09, p = .005; CI: .077/.436) and contempt (B = .28, SE = .09, p = .003; CI: .106/.467). Anger was positively albeit non-significantly related to support for
Again, to provide a stronger test of the idea that anger and contempt differentially predict normative and non-normative action, we tested whether the differences in predictive power of anger and contempt for normative and non-normative action were significant. The more restrictive model constraining the paths from contempt to normative and non-normative action to be equal differed significantly from the unconstrained model (∆χ2 = 3.93, df = 1, p < .05),
indicating that contempt differentially predicted normative and non-normative action. The results for anger also suggested differences between the prediction of normative and non-normative action; however the difference between the constrained and the unconstrained model only approached significance in this case (∆χ2 = 3.19, df = 1, p = .074).
Efficacy was unrelated to policy support (B = .08, SE = .10, p = .403; CI: -.117/.274) and negatively (but non-significantly) related to support for non-normative action (B = -.16, SE = .12, p = .196; CI: -.379/.084). Given that the two items assessing efficacy were only moderately correlated we also repeated the analysis using the single items of efficacy. We obtained a negative relation with support for violence that approached statistical significance for the item ‘Muslims can together overcome their difficulties’ (B = -.19, SE = .11, p = .073, CI: -.40/.02), but there was no relation with support for violence for the item ‘I think that Muslims as a group are able to improve their situation’ (B = -.04, SE = .10, p = .612, CI: -.24/.14). Standardized model coefficients are presented in Figure 2.
psychological distancing from the object of contempt. Thus, it seems that when contempt is felt in relation to a political issue, this emotion might be associated with a distancing from the political system and lack of endorsement for actions taken by the dominant group. As expected, anger was unrelated to non-normative action. The difference between the relations of anger with non-normative and anger with normative action approached significance, but anger was overall not significantly related to support for government policies. The latter finding is not surprising and likely to be due to the nature of the dependent measure, which assessed attitudinal support for an action taken by the government (which may be seen as an outgroup) rather than ingroup collective action. There was only limited support for our efficacy hypothesis. Although we would not expect efficacy to be related to support for government policies, which do not represent an action taken by the ingroup, the expected negative relation between efficacy and support for violence was not significant. Follow-up analyses, which yielded a negative relation between efficacy and support for non-normative action only for one of the two items, suggest that this might be due to the nature of the measure. The lack of a significant relation could also be due to larger standard errors resulting from the fact that the items were not presented in respondents’ first language. Nonetheless, an additional reason for the weak and non-significant relation between efficacy and violence support may have been that the target of violence was not specified. Thus, the imagined target might have varied across participants, such that some participants may have thought about government or military targets of violence, while others may have imagined civilian targets. It is likely that the link between low efficacy and violence support is stronger when violence is targeted at the government or other agent that is viewed as the source of oppression. Our final study examines the role of target of violence more
civilians. This study was conducted in the context of British Muslims’ responses to British foreign policy, where political participation and attitudes towards violence have been hotly debated in the public and political sphere in the last few years. The study included not just a more comprehensive range of criterion variables, but also used a measure of political efficacy rather than general group efficacy as a predictor.
Study 3
There are approximately 1.6 million Muslims in Britain, making Islam the second largest religion in the country (see Peach, 2006). British Muslims have been politically organized and active since the 1960s to address issues of racial discrimination and minority rights (see
Vertovec, 2002). Due to a series of national and international events, including the first Gulf war in 1991 and the debate around Muslim faith schools, there was a shift from racially to more religiously oriented activism in the 1980s and 1990s. This was accompanied by the emergence of Islamist groups that started to have an influence particularly on younger Muslims (see Mizra, Senthilkumaran, & Ja’far, 2007; Wiktorowicz, 2005).
Recent years have seen a further increase in the level and range of political activism in the British Muslim communities (see Briggs, 2010). This renewed interest has been stimulated to a large extent by issues surrounding social justice and by British foreign policy towards Muslim countries, in particular the Iraq war in 2003 which many Muslims (as well as many
CBS, 2006; The Independent, 2010). Some findings also suggest that a sizable minority of British Muslims felt that the 7/7 bombings were justified (e.g., Populus, 2006; GfK NOP, 2006).
The purpose of the present study was to examine the appraisals and emotions underlying support for normative and non-normative forms of political action in this context. Specifically, the current study examines British Muslims’ appraisals of British foreign policies in Muslim countries, emotions (anger and contempt) felt in relation to these policies, and efficacy perceptions as predictors of willingness to get engaged in normative collective actions and attitudes towards violence against both military and civilian targets (i.e., non-normative action). We also included a measure of participants’ willingness to vote in the next general election. Although voting intention is not directly or exclusively related to attempts to influence foreign policy, it is an intriguing dependent variable for two reasons. First of all, voting represents a contentious issue within the Muslim community, where some groups argue that voting in Britain is un-Islamic and against the teachings of Shari’a law, whereas others emphasize that voting is both a civil and religious duty and an important part of integrating into British society (see Blogspot, 2010). Secondly, since voting can be viewed as a less confrontational and system-supporting form of political action, including voting intention as an additional dependent variable allows us to test whether the negative link between contempt and support for
government policies obtained in Study 2 is generally replicable. This would provide more direct evidence that contempt is associated with disaffection with, and distancing from, the political system.
group efficacy as measured in the previous studies. Political efficacy is a theoretical concept in political science that represents citizens' faith and trust in government and the belief that they are able to influence political affairs. It therefore represents a measure of efficacy that relates more specifically to actions within the current political system. Political efficacy was shown to be strongly positively related to engagement in a variety of political actions including voting and campaign involvement. However, to our knowledge, political efficacy has not yet been examined in relation to attitudes towards political violence. Anecdotal evidence suggests, however, that in particular disenfranchised individuals who have lost trust in the government become attracted to more extremist groups (e.g., Glynn, 2002). Thus, we would again predict a negative relation between political efficacy and support for non-normative action.
Method
Participants and Procedure
The study was administered as an online survey. Respondents were recruited using an advertisement on Facebook which targeted Facebook users living in the United Kingdom and aged 18 years or older. To ensure that the advertisement would reach many Muslims, it targeted users who had words related to Islam or Muslims (e.g., Islam, Muslims, Arabic, Bangladesh) in their profile. The advertisement depicted the British flag and the flag of the Muslim league and the question ‘Are you Muslim?’. When clicking on the link respondents were taken to the survey. Respondents were given the opportunity to enter into a prize draw at the end of the survey. The survey was completed by 473 respondents. Seven respondents who specified that they were not Muslims were excluded. Our final sample comprised 466 Muslims living in Britain (247 female, 215 male, 4 unknown; mean age = 26.69, SD = 8.10).
In addition to a number of measures relating to issues affecting Muslims in Britain, the questionnaire contained items assessing our key constructs.
Injustice appraisals
Appraisals of (in-)justice of British foreign policies were measured by six items. On scales ranging from 1 (not at all) to 7 (extremely), respondents indicated the extent to which they felt that British foreign policy in the Middle East, Britain’s role in the invasion of Iraq in 2003, and Britain’s current campaign against the Taliban in Afghanistan were illegitimate and immoral (α = .93).
Political efficacy
The political efficacy items were adapted from established scales (Campbell et al., 1954; Craig et al., 1990; Muller & Jukam, 1977). On scales ranging from 1 (strongly disagree) to 7 (strongly agree), respondents indicated their agreement with the following items: “The way people vote is the main thing that decides how things are run in this country”; “I feel that I am quite well represented in our political system”; “In general, I can rely on the government to do the right thing”. The items were averaged to yield an index of political efficacy (α = .64). Anger
On scales ranging from 1 (not at all) to 7 (extremely), respondents indicated the extent to which they felt anger, outrage, and frustration when thinking about British foreign policy towards Muslim countries in the recent past (α = .92).
Contempt
Normative collective action intention
Respondents indicated how willing they were to engage in the following actions to change British foreign policy towards Muslim countries: sign a petition to the government, join in a peaceful public rally, protest, or demonstration in support of Muslims, and lobby an MP (1 = not at all willing; 9 = very willing). The items were averaged to yield an index of normative collective action intentions (α = .80).
Voting intention
Using a scale from 1 (not at all likely) to 7 (very likely) respondents indicated how likely they thought they were to vote in the next general election.
Support for violence
Attitudes towards violence were measured by seven items related to both attacks against military forces in Muslim countries and attacks against civilians in Western countries.
Respondents indicated the extent to which they can understand the reasons why some groups might resort to violence to force Western military forces out of Muslim countries (0 = not at all; 9 = very much), the extent to which they felt that it is justified for groups to use violence to force Western military forces out of Muslim countries (0 = absolutely unjustified; 9 = absolutely justified), the extent to which they support or oppose violence by Islamist groups against
might have wanted to carry out suicide operations in Britain (0 = not at all; 9 = very much), and the extent to which they felt the 2005 London bombings were justified or unjustified (0 = absolutely unjustified; 9 = absolutely justified).
Principal components analysis of these items yielded two components with eigenvalues greater than 1 which accounted for 62.46 % of the variance. Loadings, after oblique rotation, indicated that items relating to violence against military targets (understanding violence against Western military targets, violence against military targets justified, support for violence against military targets, opinion of British Muslim fighters) loaded on the first component (>.48) and items relating to violence against civilian targets (support for violence against civilian targets, understanding British Muslims wanting to carry out suicide bombings in the UK, 7/7 bombings justified) loaded on the second component (>.47). The items were standardized and then averaged to yield indices of attitudes towards violence against military targets (α = .80) and attitudes towards violence against civilian targets (α = .64). Means of unstandardized items are presented in Table 3. The index of attitudes towards violence against civilians deviated from normality (skewness = 2.07, kurtosis = 4.99) and was transformed with a logarithmic
transformation before our analyses (skewness = .67, kurtosis = -.46). Results and Discussion
Means, standard deviations, and zero-order correlations between key variables are presented in Table 4. Again we performed regression analyses in Mplus to test our hypotheses. Because there was a small amount of scattered missing data (≤ 7.7%), we again used FIML estimation.
political efficacy negatively predicted anger (B = -.20, SE = .06, p < .001; CI: -.310/-.092) but was not significantly related to contempt (B = -.12, SE = .07, p = .102; CI: -.265/.025). The negative relation between political efficacy and anger was unexpected, but is not surprising. It suggests that individuals who have low political efficacy (e.g., who do not trust the government to do the right thing and feel that they do not have the ability to influence policy decisions) respond with more anger to British foreign policy decisions. Anger significantly predicted willingness to engage in normative collective action (B = .46, SE = .08, p < .001; CI: .305/.609) as well as attitudes towards violence against military targets (B = .12, SE = .03, p < .001; CI: .057/.173). Anger was unrelated to voting intention (B = .12, SE = .09, p = .162; CI: -.047/.285) and attitudes towards violence against civilians (B = .01, SE = .01, p = .203; CI: -.007/.033). Contempt positively predicted both attitudes toward violence against military targets (B = .06, SE = .02, p = .021; CI: .009/.100) and attitudes toward violence against civilian targets (B = .02, SE = .01, p = .038; CI: .001/.034). Contempt was unrelated to normative collective action intention (B = -.01, SE = .06, p = .905; CI: -.137/.118) and negatively (although only approaching
statistical significance) related to voting intention (B = -.13, SE = .07, p = .074; CI: -.270/.009). There was also a direct relation between injustice perceptions and normative collective action intention (B = .18, SE = .06, p = .005; CI: .051/.300).
normative and on the two types of non-normative action, respectively. This model differed (although only approaching significance) from a more restrictive model where all four paths from contempt to action were constrained to be equal (∆χ2 = 3.58, df = 1, p = .06), suggesting again that contempt differentially predicts normative and non-normative action. For anger, the model constraining the relations for anger and the two normative actions and anger and the two non-normative actions differed significantly from an unconstrained model (∆χ2 = 22.86, df = 2, p < .001), indicating that anger had different effects within each type of action (normative and non-normative). We therefore conducted pair wise comparisons of the effects of anger for each of the actions. These comparisons indicated that anger was significantly more strongly related to normative collective action than to any of the other forms of action (all ps < .001) and also that anger was more strongly related to support for violence against military targets than to violence against civilian targets (∆χ2 = 16.13, df = 1, p < .001). There were no significant differences between the relation between anger and voting and anger and the two types of violence support.
Political efficacy positively predicted voting intention (B = .25, SE = .08, p = .003; CI: .081/.406) and collective action intention (B = .16, SE = .08, p = .050; CI: -.008/.313) and was negatively related to attitudes towards violence against military targets (B = -.06, SE = .03, p = .016; CI: -.112/-.013). Political efficacy was, however, unrelated to attitudes towards violence against civilian targets (B = -.01, SE = .01, p = .619; CI: -.025/.014). Standardized coefficients are presented in Figure 3.
violence against military targets (B = .02, SE = .01, p = .037; CI: .003/.036) and attitudes towards violence against civilian targets (B = .01, SE = .00, p = .058; CI: .001/.012) via contempt. There were also significant indirect effects of political efficacy on collective action intention (B = -.09, SE = .03, p = .006; CI: .165/.040) and attitudes towards violence against military targets (B = -.02, SE = .01, p = .009; CI: -.046/-.009) via anger.
foreign policy. Although this relation was not specifically hypothesized, it seems plausible given the nature of the efficacy measure. If people feel they are well-represented in the political system and that the system is responsive to their concerns, they have less reason to feel angry about policy decisions.
Overall, the present study provided additional support for our theoretical ideas in a context of a disenfranchised group, considering a wider range of forms of political engagement, and using an index of political efficacy which represents the likely impact of actions within the current political system more specifically. This study therefore further supports the
generalizability and robustness of our theoretical ideas. General Discussion
The purpose of this research was to expand on existing work on collective action, which has paid relatively little attention to the factors underlying more radical, non-normative group-based actions. To this end we extended a recent integrative model (van Zomeren et al., 2004) and tested several novel ideas regarding the relations between emotion, efficacy and normative and non-normative action, across three diverse political contexts. In the following sections we will first evaluate our results in relation to our key predictions and suggest directions for further research on these ideas. We will then draw attention to a number of limitations of our research and finally highlight theoretical contributions and practical implications of our findings.
The Role of Emotions in Normative and Non-Normative Collective Action: Anger vs. Contempt Our first main idea suggested that qualitatively different emotions would predict
action. We also hypothesized that both these emotions would be predicted by appraisals of injustice, and that injustice perceptions would therefore be indirectly linked to normative and non-normative collective action via anger and contempt, respectively.
Consistent with our hypothesis, and with previous research on the role of anger in collective action (see van Zomeren et al., 2008), anger emerged as a significant predictor of willingness to engage in normative collective action both in the context of student protests in Germany (Study 1) and British Muslims’ responses to British foreign policy (Study 3). Study 2 suggested, however, that anger is not a significant predictor of support for normative action taken by another agent (in this case the government). Rather, support for normative action was directly predicted by injustice appraisals in this case. This suggests that anger might be particularly pertinent in motivating own action or support for ingroup action and less predictive if it comes to attitudinal support for action taken by other agents or outgroups.
also on the surface seem as conflicting with our findings. This work has, however, focused more on the negative arousal associated with in situ anger and frustration and the impulsive aggressive behaviours that follow, rather than the relation between the appraisal component of anger and instrumental forms of action examined here.
Our findings are, however, wholly in line with a view of anger as a constructive emotion that occurs in close relationships and functions to correct wrongdoing and uphold moral
standards (see Averill, 1983; Fischer & Roseman, 2007; Weber, 2004). This work has also shown that anger mostly results in actions that have beneficial consequences for social relations and make reconciliation possible (Averill, 1983; Fischer & Roseman, 2007; Weber, 2004). In the context of political action, this means that people who feel angry about an unjust treatment still feel connected to the political system and therefore are more likely to engage in and support action within the confines of the system.
less normative. This was evident in both Study 1 and Study 3. In Study 1 anger is most predictive of normative action, moderately predictive of non-violent, non-normative action such as
blocking streets and buildings, and least predictive of violent action. Similarly in Study 3 anger most strongly predicts willingness to engage in normative action, less strongly predicts support for violence against military targets, and least strongly support for violence against civilians (Anger was unrelated to voting but this was expected as voting did not relate directly to the injustice). Future research may examine role of anger in predicting different forms of collective action further, investigating, for example, whether different forms of anger (see Russell & Fehr, 1994) have differential implications for different forms of action.
The findings regarding our hypothesis that contempt would be a predictor of non-normative action were highly consistent across the three studies. In line with our predictions, feelings of contempt positively predicted likelihood to engage in violent non-normative action in the context of German student protests (Study 1), support for political violence among Indian Muslims (Study 2) and support for violence against both military and civilian targets among British Muslims (Study 3). To our knowledge, the present research is the first to provide
evidence for this link. These findings are generally in line with the suggestion that contempt may be associated with particularly hostile reactions because attack tendencies are not held in check by a desire to preserve social relationships (Fischer & Roseman 2007), and because the
accompanying derogation and moral exclusion can serve to legitimize extreme actions against an offender. The results are also consistent with the idea that contempt, which implies a
that contempt is negatively related to actions that might be seen as system-supporting (support for government policies and voting). Thus, feelings of contempt in a political context might signal disaffection from the political system more generally. These results underline the
importance of examining this emotion as a determinant of various forms of political (in)action. As hypothesized, both anger and contempt were predicted by perceptions of injustice, and injustice appraisals were significantly indirectly related to normative and non-normative
collective action via these emotions. While the role of injustice appraisals is central in appraisal theories of emotion (e.g., Frijda et al., 1989) and models of political action (e.g., Gurr, 1970; Runciman, 1966; Tajfel & Turner, 1979) and terrorism (Moghaddam, 2005), we believe that it is imperative that future research further investigates the contextual and psychological factors that determine whether injustice appraisals result in anger or contempt. Some research implies that anger and contempt result from different forms of norm violation (e.g., violations of autonomy vs. community, respectively; Rozin, Lowery, Imada, & Haidt, 1999), while other work suggests that anger and contempt can result from the same incidences. Fischer and Roseman (2007) showed for example that contempt often arises when prior incidents of anger went unresolved and there is a perceived lack of control over the other person. In the domain of group-based injustices and political action, this suggests that contempt may evolve when previous attempts to address an injustice were futile. Repeated violations of human moral standards by the
government or other powerful group, such as discounting numerous civilian casualties as ‘collateral damage’ in a conflict, are also likely to provide a fertile ground for the development of (political) contempt.
value placed on, relationships or identities might determine whether people respond to an injustice with anger or contempt. Thus, contempt may be preceded or accompanied by disidentification from the political system or disaffection from society more generally and a sense of belonging to, or identification with, a superordinate group or political entity might determine whether an experienced injustice is responded to with anger or contempt.
Another intriguing possibility is that there are individual differences in the propensity to respond with contempt and the willingness to engage in violent, hostile behaviour. Recent research has demonstrated that individual differences in responding with particular emotions partly explain prejudicial reactions to certain outgroups (see Hodson & Costello, 2007). Future research could explore whether such individual propensities also play a role in determining emotional reactions to political events and willingness to engage in different forms of political action.
This finding is significant in that is goes against traditional thinking in the literature that collective action primarily happens in unstable social systems (Tajfel & Turner, 1979) and is driven by high efficacy beliefs (van Zomeren et al., 2008). It is, however, consistent with recent theoretical extensions of SIT that suggest that more confrontational, non-normative action strategies are chosen under desperate conditions, such as when low status is stable (Scheepers et al., 2006; Spears et al., 2010) or when legitimate channels to achieve social change are closed (Wright et al., 1990a). This finding also resonates with the terrorism literature which stresses the role of powerlessness against state power in driving such action (e.g., Moghaddam, 2005).
As we have suggested earlier, the fact that low group efficacy predicts non-normative action does not imply that non-normative action is an irrational strategy. Non-normative action can be highly strategic, and can fulfil a number of short-term goals that contribute towards achieving the desired social change in the long run (see Hornsey at al., 2006). For example, in an analysis of Al-Qaeda strategy, Sedgwick (2004) suggests that the purpose of the 9/11 attacks was to provoke a counter-attack from the US that would then have a radicalizing effect on
al., 1990a), is marginalized by the procedures of the existing political system (Gurr, 1993; Schwarzmantel, 2010), or is too disorganized or unsupportive of the cause to bring about the mass action required to effect change with means within the system (see van Zomeren et al., 2004). Thus, it is possible that our measures of group efficacy in Studies 1 and 2 may have evoked normative action strategies. Study 3, which examined political efficacy, a concept that refers more specifically to the efficacy of actions within the system, suggests that the efficacy of normative actions is negatively related to support for violence. We suggest that future work should additionally examine the efficacy of violent actions and explore the interactions between the perceived efficacy of different forms of action. We would like to point out, however, that this does not mean that general group efficacy is irrelevant as a predictor. As our results show, there is a meaningful negative relation between group efficacy and non-normative action, indicating that the more general collective strength a group has, the less likely it will need to resort to extreme or violent measures. While efficacy measured generally may well be associated with the efficacy of normative action, and the efficacy of non-normative action might be a more specific and positive predictor, a key psychological point is that these two forms of efficacy will often be negatively related. Future work might therefore also examine the interplay of general group efficacy and specific forms of action efficacy.
suggest that future research also takes into account the perceived efficacy of specific politicized subgroups or vanguard groups. It is likely that individuals who view the ingroup as a whole (e.g., Muslims) as weak and ineffective might become involved in non-normative activities when they view a certain vanguard group (e.g., al-Qaeda) as strong and effective in redressing group-based injustices (see Husain, 2007). In fact, militant extremist groups often portray their groups as the only effective agent to bring about the desired change (see Saucier, Akers, Shen-Miller,
Knezevic, & Stankov, 2009). Thus it is likely that it is the perceived efficacy of such militant subgroups that makes these groups attractive to disaffected individuals (see also our discussion on the role of identification below). Thus, we would expect a positive relation between the efficacy of such militant subgroups and support for, and engagement in, non-normative action (see also Louis, 2009, for similar arguments).
Limitations of the Present Research
We acknowledge several limitations of our studies. The focus of our analyses was on the predictive roles of several theoretically relevant variables and our regression approach allowed us to isolate the contributions of these variables in predicting normative and non-normative action support. It is important to note, however, that our reliance on cross-sectional data
(Weis et al., 1999). Nonetheless, the causal relations between variables are also likely to be reciprocal and some variables such as emotions and action tendencies are likely to arise more or less simultaneously, as suggested by appraisal theories of emotions (e.g., Frijda et al., 1989). Thus, conceptualizing a strict causal order between variables might be counterproductive. However, we can of course not preclude the possibility that, for example, non-normative action tendencies or support are legitimized by invoking the injustice of a situation, the low efficacy of the group in achieving the desired social change, and the contempt-worthiness of the opponent. It is therefore imperative that future longitudinal and experimental work corroborates our findings.
Furthermore, like most research on collective action (see van Zomeren et al., 2008), we relied on behavioural intentions and action support as our criterion variables. Although previous research has shown that behavioural intentions are a proxy for actual behaviour and predict actual participation in collective action (e.g., De Weerd & Klandermans, 1999; Moskalenko & McCauley, 2009; Webb & Sheeran, 2006), and that attitudes towards different forms of action in a community can predict actual action such as terrorism (Krueger & Malečková, 2009), it would be desirable that future research further strengthens the present findings by investigating actual participation in normative and non-normative collective action. Based on previous findings we would expect similar (albeit potentially smaller; see van Zomeren et al., 2008) relations between our explanatory variables and actual participation.
permeability; Tajfel & Turner, 1979) or a salient social identity, were assumed rather than specifically assessed and integrated into our analysis. It is therefore likely that other variables will further contribute, mediate, or moderate the relations tested in the present work. For example, identification with a disadvantaged group is likely to be a distal predictor, such that, individuals who identify more strongly with their group would be more likely to appraise an event that disadvantages the ingroup as unjust (see van Zomeren et al., 2008). Identification could also moderate the link between injustice appraisals and emotions, such that high identifiers react more strongly emotionally to injustices committed against the ingroup. Moreover, the presence of an injustice can politicize identities (see Simon & Klandermans, 2001), and identification with a politicized subgroup is likely to mediate the relations between our
explanatory variables and forms of collective action (see Stürmer & Simon, 2004). For example, individuals who perceive the ingroup as ineffective or who have lost faith in, and feel contempt for, the political system might become attracted to politicized groups who provide an ideology that undermines the dominant system and legitimizes the use of non-normative means to redress group-based injustices, for example by dehumanizing the enemy outgroup and presenting the ingroup as virtuous and as fighting the spread of evil (Saucier et al., 2009). Identification with such groups might then proximally predict engagement in non-normative action.
Contributions and Implications of the Present Research
